<p>The current study has investigated two-phase refrigerant flow distribution in heat exchange manifolds. Experimental data have been acquired in a heat exchanger test rig specially made for measurement of mass flow rate and gas and liquid distribution in the manifolds of compact heat exchangers. Twelve different manifold designs were used in the experiments, and CO² and HFC-134a were used as refrigerants. </p>

<p>The presence of deposits in the honing grooves in diesel engine cylinder liners can cause a severe increase in the consumption of lubricating oil. These deposits may appear amber in which case they may be referred to as 'liner lacquerer' or be nearly invisible in which case they are often referred to as 'transparent glaze' or 'bore glaze' in the marine sector. The formation of these deposits is believed to be influenced by engine design, engine load, fuel composition and lubricating oil composition. This relationship has, however, not been well understood and little material has been published on this subject. This thesis describes an investigation of this phenomenon. The problem has been approached by studying the composition of deposit samples, studying cases of deposit formation in the field and through experimental work.</p><p>As a result of this work, it is hypothesized that the root cause of the deposit formation is a mismatch between the rate of formation of oil insoluble material on or within the oil film and the oil film dispersing power and exchange rate. A large number of parameters will affect this balance which explains the sliding shift in appearance and composition of 'liner lacquer' and 'bore glaze' and the difficulty in identifying the cause of this problem.</p><p>A combined mass balance and chemical kinetics approach is used to bridge the gap between fundamental deposit theory and tangible engine related parameters. It is thus possible to rationalize the formation of deposits on cylinder liners. This understanding is sufficient to point out which factors should be considered in terms of the prevention of deposit formation and to present a viable hypothesis on the reason for the deposit formation in the engines that have been investigated in the course of this work as summarized in the following paragraph.</p><p>The presence of deposit indicates that the dispersing power and transport rate of the oil film is insufficient to deal with the deposit precursors being formed. This is believed related to extensive low load operation which is associated with both low liner temperatures and low nominal oil consumption in these engines. Low liner temperatures will encourage the formation of insoluble material due to condensation of sulphuric acid precursors, while low oil consumption is believed to indicate low oil film exchange which reduces the transport of matter and also contributes to oil film oxidation by prolonging the exposure to combustion gases.</p>

<p>Frequency and amplitude dependency of velocity distribution and head loss of oscillatory flow in a rectangular duct have been studied experimentally with a model tunnel system.</p><p>Tests were carried out with the duct of both smooth and rough walls. The smooth wall was made of Plexiglas. Sand roughness was used for the rough wall. Velocity, pressure and differential pressure of stationary flow, pure oscillatory flow and combined oscillatory flow were measured. The combined oscillatory flow was classified as oscillation dominant flow, stationary dominant flow, and oscillation-and-stationary balanced flow. Various oscillating frequencies, amplitudes and steady flow percentages were tested for oscillatory flows. The oscillating frequencies tested were varied from 0.01 Hz to 1.00 Hz. Oscillatory amplitude and stationary part were varied from 10 to 100%. Velocity of the flow was measured with a 2D PIV (Particle Image Velocimetry) and a 2D LDV (Laser Doppler Velocimetry) respectively at different test stages. The maximum mainstream velocity was ranged from 0.05 m/s to 1.1 m/s. Data of pressure variations along the tunnel were collected with differential pressure sensors. Flow rate and instant wall pressures at multiple points along the test tunnel were measured simultaneously. The static pressure in the test tunnel was about 1.0 mWC. The differential pressure along the tunnel was less than 20 mmWC per meter.</p><p>Examples of velocity distribution in the test rig from LDV measurement are presented, for both stationary flow and oscillatory flow. The dimensionless velocity distributions of stationary flow are in good agreement with the universal velocity distribution law. Deviations are obvious between the velocity distributions of oscillatory flow and the universal velocity distribution law, when the measured velocity is scaled to dimensionless by friction velocity from Clauser chart. Examples of PIV velocities of different flow regimes are presented in the forms of velocity profile and velocity waveform. Generally, the velocity distributions are in good agreement with the results from LDV, in agreement with the normal turbulent velocity distribution in a duct, if the velocity magnitude is not too small. Dimensionless velocity profiles at various phase angles of the same oscillatory flow regime have quite consistent distribution. The annular effect is observed in some cases. Its occurrence depends on the complex actions of oscillating frequency, amplitude and stationary flow percentage. The velocity waveform confirms the characteristics of mass oscillation of the flow. No significant phase shift is displayed between the velocity waveform of the boundary and centreline in most cases tested. The vertical velocity, which is normal to the mainstream, is quite small and has similar features to the mainstream velocity.</p><p>The velocity profiles got from both LDV and PIV show that the flows in the test tunnel were typical turbulence, with typical velocity distribution of turbulence. No transition between laminar and turbulence is observed even at the turning point of oscillation.</p><p>Pressure variations measured along the tunnel of different flow regimes are presented. The accelerative heads of oscillatory flow are calculated. The friction head losses along the tunnel are evaluated. The dependencies of pressure variation and friction head loss on oscillatory frequency and amplitude are investigated for both pure oscillatory flow and combined oscillatory flow. It is proven that the friction head loss of oscillatory flow increases along with the increase of frequency if the mean flow rate is kept constant. The peak friction head loss increases along with the increase of oscillatory amplitude. Comparison of pressure variation and friction head loss between stationary flow and oscillatory flow shows that the friction head loss of unsteady flow is much bigger than that of steady flow. This is in good agreement with the expectancy based on the experimental results of laminar flow. The head loss of pure oscillatory flow was greater than that of the stationary flow for dozens or more times for various flow regimes running at equivalent flow rate. The ratio of head loss of combined oscillatory flow to stationary flow running at equivalent flow rate is smaller than that of pure oscillatory flow to stationary flow, several times to dozen times. In general, the frequency dependency of head loss on oscillatory frequency and amplitude is clear, though the measuring length is only 9 metres and the absolute magnitude of pressure variation is less than 0.20 mWC.</p>

<p>This thesis focuses on three main areas within the field of SOFC/GT-technology:</p><p>• Development of a dynamic SOFC/GT model</p><p>• Model calibration and sensitivity study</p><p>• Assessment of the dynamic properties of a SOFC/GT power plant</p><p>The SOFC/GT model developed in this thesis describes a pressurised tubular Siemens Westinghouse-type SOFC, which is integrated in a gas turbine cycle. The process further includes a plate-fin recuperator for stack air preheating, a prereformer, an anode exhaust gas recycling loop for steam/carbon-ratio control, an afterburner and a shell-tube heat exchanger for air preheating. The fuel cell tube, the recuperator and the shell-tube heat exchanger are spatially distributed models. The SOFC model is further thermally integrated with the prereformer. The compressor and turbine models are based on performance maps as a general representation of the characteristics. In addition, a shaft model which incorporates moment of inertia is included to account for gas turbine transients.</p><p>The SOFC model is calibrated against experimentally obtained data from a single-cell experiment performed on a Siemens Westinghouse tubular SOFC. The agreement between the model and the experimental results is good. The sensitivity study revealed that the degree of prereforming is of great importance with respect to the axial temperature distribution of the fuel cell.</p><p>Types of malfunctions are discussed prior to the dynamic behaviour study. The dynamic study of the SOFC/GT process is performed by simulating small and large load changes according to three different strategies;</p><p>• Load change at constant mean fuel cell temperature</p><p>• Load change at constant turbine inlet temperature</p><p>• Load change at constant shaft speed</p><p>Of these three strategies, the constant mean fuel cell temperature strategy appears to be the most rapid load change method. Furthermore, this strategy implies the lowest degree of thermal cycling, the smoothest fuel cell temperature distribution and the lowest current density at part-load. Thus, this strategy represents the overall lowest risk with respect to system malfunctions and degradation. In addition, the constant mean fuel cell temperature strategy facilitates high efficiency part-load operation. The constant turbine inlet temperature strategy proved to lead to unstable operation at low load, and thus it is considered to be the least adequate method for load change. For both the constant mean fuel cell temperature strategy and the constant TIT strategy, surge might be a problem for very large load reductions. The slowest response to load changes was found for the constant shaft speed strategy. Furthermore, this strategy leads to very low fuel cell temperatures at low loads. This in combination with a possible higher degradation rate makes the constant shaft speed strategy unsuited for large load variations. Nevertheless, operation at constant shaft speed may be facilitated by air bypass, VIGV or compressor blow off.</p> / Paper I is published with kind permission of Elsevier, Sciencedirect.com

The presence of deposits in the honing grooves in diesel engine cylinder liners can cause a severe increase in the consumption of lubricating oil. These deposits may appear amber in which case they may be referred to as 'liner lacquerer' or be nearly invisible in which case they are often referred to as 'transparent glaze' or 'bore glaze' in the marine sector. The formation of these deposits is believed to be influenced by engine design, engine load, fuel composition and lubricating oil composition. This relationship has, however, not been well understood and little material has been published on this subject. This thesis describes an investigation of this phenomenon. The problem has been approached by studying the composition of deposit samples, studying cases of deposit formation in the field and through experimental work. As a result of this work, it is hypothesized that the root cause of the deposit formation is a mismatch between the rate of formation of oil insoluble material on or within the oil film and the oil film dispersing power and exchange rate. A large number of parameters will affect this balance which explains the sliding shift in appearance and composition of 'liner lacquer' and 'bore glaze' and the difficulty in identifying the cause of this problem. A combined mass balance and chemical kinetics approach is used to bridge the gap between fundamental deposit theory and tangible engine related parameters. It is thus possible to rationalize the formation of deposits on cylinder liners. This understanding is sufficient to point out which factors should be considered in terms of the prevention of deposit formation and to present a viable hypothesis on the reason for the deposit formation in the engines that have been investigated in the course of this work as summarized in the following paragraph. The presence of deposit indicates that the dispersing power and transport rate of the oil film is insufficient to deal with the deposit precursors being formed. This is believed related to extensive low load operation which is associated with both low liner temperatures and low nominal oil consumption in these engines. Low liner temperatures will encourage the formation of insoluble material due to condensation of sulphuric acid precursors, while low oil consumption is believed to indicate low oil film exchange which reduces the transport of matter and also contributes to oil film oxidation by prolonging the exposure to combustion gases.

The current study has investigated two-phase refrigerant flow distribution in heat exchange manifolds. Experimental data have been acquired in a heat exchanger test rig specially made for measurement of mass flow rate and gas and liquid distribution in the manifolds of compact heat exchangers. Twelve different manifold designs were used in the experiments, and CO2 and HFC-134a were used as refrigerants.

Frequency and amplitude dependency of velocity distribution and head loss of oscillatory flow in a rectangular duct have been studied experimentally with a model tunnel system. Tests were carried out with the duct of both smooth and rough walls. The smooth wall was made of Plexiglas. Sand roughness was used for the rough wall. Velocity, pressure and differential pressure of stationary flow, pure oscillatory flow and combined oscillatory flow were measured. The combined oscillatory flow was classified as oscillation dominant flow, stationary dominant flow, and oscillation-and-stationary balanced flow. Various oscillating frequencies, amplitudes and steady flow percentages were tested for oscillatory flows. The oscillating frequencies tested were varied from 0.01 Hz to 1.00 Hz. Oscillatory amplitude and stationary part were varied from 10 to 100%. Velocity of the flow was measured with a 2D PIV (Particle Image Velocimetry) and a 2D LDV (Laser Doppler Velocimetry) respectively at different test stages. The maximum mainstream velocity was ranged from 0.05 m/s to 1.1 m/s. Data of pressure variations along the tunnel were collected with differential pressure sensors. Flow rate and instant wall pressures at multiple points along the test tunnel were measured simultaneously. The static pressure in the test tunnel was about 1.0 mWC. The differential pressure along the tunnel was less than 20 mmWC per meter. Examples of velocity distribution in the test rig from LDV measurement are presented, for both stationary flow and oscillatory flow. The dimensionless velocity distributions of stationary flow are in good agreement with the universal velocity distribution law. Deviations are obvious between the velocity distributions of oscillatory flow and the universal velocity distribution law, when the measured velocity is scaled to dimensionless by friction velocity from Clauser chart. Examples of PIV velocities of different flow regimes are presented in the forms of velocity profile and velocity waveform. Generally, the velocity distributions are in good agreement with the results from LDV, in agreement with the normal turbulent velocity distribution in a duct, if the velocity magnitude is not too small. Dimensionless velocity profiles at various phase angles of the same oscillatory flow regime have quite consistent distribution. The annular effect is observed in some cases. Its occurrence depends on the complex actions of oscillating frequency, amplitude and stationary flow percentage. The velocity waveform confirms the characteristics of mass oscillation of the flow. No significant phase shift is displayed between the velocity waveform of the boundary and centreline in most cases tested. The vertical velocity, which is normal to the mainstream, is quite small and has similar features to the mainstream velocity. The velocity profiles got from both LDV and PIV show that the flows in the test tunnel were typical turbulence, with typical velocity distribution of turbulence. No transition between laminar and turbulence is observed even at the turning point of oscillation. Pressure variations measured along the tunnel of different flow regimes are presented. The accelerative heads of oscillatory flow are calculated. The friction head losses along the tunnel are evaluated. The dependencies of pressure variation and friction head loss on oscillatory frequency and amplitude are investigated for both pure oscillatory flow and combined oscillatory flow. It is proven that the friction head loss of oscillatory flow increases along with the increase of frequency if the mean flow rate is kept constant. The peak friction head loss increases along with the increase of oscillatory amplitude. Comparison of pressure variation and friction head loss between stationary flow and oscillatory flow shows that the friction head loss of unsteady flow is much bigger than that of steady flow. This is in good agreement with the expectancy based on the experimental results of laminar flow. The head loss of pure oscillatory flow was greater than that of the stationary flow for dozens or more times for various flow regimes running at equivalent flow rate. The ratio of head loss of combined oscillatory flow to stationary flow running at equivalent flow rate is smaller than that of pure oscillatory flow to stationary flow, several times to dozen times. In general, the frequency dependency of head loss on oscillatory frequency and amplitude is clear, though the measuring length is only 9 metres and the absolute magnitude of pressure variation is less than 0.20 mWC.

This thesis focuses on three main areas within the field of SOFC/GT-technology: • Development of a dynamic SOFC/GT model • Model calibration and sensitivity study • Assessment of the dynamic properties of a SOFC/GT power plant The SOFC/GT model developed in this thesis describes a pressurised tubular Siemens Westinghouse-type SOFC, which is integrated in a gas turbine cycle. The process further includes a plate-fin recuperator for stack air preheating, a prereformer, an anode exhaust gas recycling loop for steam/carbon-ratio control, an afterburner and a shell-tube heat exchanger for air preheating. The fuel cell tube, the recuperator and the shell-tube heat exchanger are spatially distributed models. The SOFC model is further thermally integrated with the prereformer. The compressor and turbine models are based on performance maps as a general representation of the characteristics. In addition, a shaft model which incorporates moment of inertia is included to account for gas turbine transients. The SOFC model is calibrated against experimentally obtained data from a single-cell experiment performed on a Siemens Westinghouse tubular SOFC. The agreement between the model and the experimental results is good. The sensitivity study revealed that the degree of prereforming is of great importance with respect to the axial temperature distribution of the fuel cell. Types of malfunctions are discussed prior to the dynamic behaviour study. The dynamic study of the SOFC/GT process is performed by simulating small and large load changes according to three different strategies; • Load change at constant mean fuel cell temperature • Load change at constant turbine inlet temperature • Load change at constant shaft speed Of these three strategies, the constant mean fuel cell temperature strategy appears to be the most rapid load change method. Furthermore, this strategy implies the lowest degree of thermal cycling, the smoothest fuel cell temperature distribution and the lowest current density at part-load. Thus, this strategy represents the overall lowest risk with respect to system malfunctions and degradation. In addition, the constant mean fuel cell temperature strategy facilitates high efficiency part-load operation. The constant turbine inlet temperature strategy proved to lead to unstable operation at low load, and thus it is considered to be the least adequate method for load change. For both the constant mean fuel cell temperature strategy and the constant TIT strategy, surge might be a problem for very large load reductions. The slowest response to load changes was found for the constant shaft speed strategy. Furthermore, this strategy leads to very low fuel cell temperatures at low loads. This in combination with a possible higher degradation rate makes the constant shaft speed strategy unsuited for large load variations. Nevertheless, operation at constant shaft speed may be facilitated by air bypass, VIGV or compressor blow off. / Paper I is published with kind permission of Elsevier, Sciencedirect.com

<p>Soot and radiation play an important role when designing practical combustion devices, and great efforts have been put into developing models which describe soot formation and oxidation. The Eddy Dissipation Concept (EDC) has proven to describe turbulent combustion well, and has the flexibility to describe chemical kinetics in a detailed manner. The aim of this work is to study how the EDC handles soot models based on a detailed representation of the gas-phase chemical kinetics.</p><p>Two versions of a semi-empirical soot model is used in conjunction with the EDC. Concentrations of various intermediate species are used as input to the soot models.</p><p>The implementation of the new soot models is discussed in relation to the previous implementation of a less detailed soot model. To assure that the interaction between soot and the gas-phase species is represented correctly, the soot models are implemented with a two-way coupling of soot and gas-phase kinetics.</p><p>Soot is a good radiator. In a sooting flame a substantial amount of energy will be transferred to the surroundings by thermal radiation. This transfer of energy will alter the temperature field of the flame and the change in temperature will affect the kinetics of soot and gas-phase chemistry. To simulate sooting flames correctly, it was therefore necessary to include a radiation model.</p><p>To validate the coupled models of turbulence, combustion, soot, and radiation two different turbulent flames were simulated. One turbulent jet flame of methane and one turbulent jet flame of ethylene. For both flames the computed results were compared with measured values.</p><p>Several aspects of the simulations are studied and discussed, such as the effect of the two-way coupling of soot and gas-phase kinetics on both soot yield and gas-phase composition, and the importance of a suitable radiation model.</p><p>The two-way coupling of soot and gas phase kinetics is shown to have a positive effect on the computed soot volume fractions, and the results are considered to be encouraging. The work has demonstrated that the EDC has the capacity to handle different types of chemical reaction mechanisms, such as mechanisms for gas-phase combustion and soot kinetics, without modification.</p>

<p>In this thesis the evaporation from and condensation on a plane liquid surface have been studied by analysis and molecular dynamics simulations. The effect of the condensation coefficient on the inverted temperature gradient for a two-surface evaporation-condensation geometry is investigated by the moment method. The influence of the molecular exchange phenomenon on the gas-kinetic treatment of evaporation and condensation is shown to be neglible under certain assumptions. Methods to simulate half-space steady evaporation or condensation in Direct Simulation Monte Carlo simulations are adapted to Molecular Dynamics (MD). A microscopic definition of evaporation and condensation is introduced and values for the evaporation and condensation coefficients are calculated from MD. The velocity distribution functions for the evaporation and condensation modes have been calculated and compared with the standard assumptions in gas-kinetic calculations. </p>